Full Idea
Duhem and Quine have maintained that it may be possible to develop two or more theories that are 1) internally consistent, 2) inconsistent with one another, and 3) perfectly consistent with all the data we can muster.
Gist of Idea
Two theories can be internally consistent and match all the facts, yet be inconsistent with one another
Source
report of Willard Quine (Word and Object [1960]) by J Baggini / PS Fosl - The Philosopher's Toolkit §1.06
Book Reference
Baggini,J and Fosl,P.S.: 'The Philosopher's Toolkit' [Blackwells 2003], p.20
A Reaction
Obviously this may be a contingent truth about our theories, but why not presume that this is because we are unable to collect the crucial data (e.g. about prehistoric biology), rather than denigrate the whole concept of a theory, and undermine science?